OpenAI and Oracle todayAnnounced a Binding Agreement for the Stargate NetworkThe addition of 4.5 gigawatts of high-power-density data center capacity moves the project past the 5-gigawatt milestone and enables it to house more than 2 million sets of AI chips.
While SoftBank was still a name in Stargate's January press release, the Japanese conglomerate did not contribute to that specific funding, so Oracle is now the primary financier and operator.
The new expansion will connect to Stargate I in Abilene, where Oracle delivered the first NVIDIA GB200 racks last month and where OpenAI already runs cutting-edge model training workloads. Construction crews from more than 20 states are on site, and OpenAI expects the expanded footprint will create more than 100,000 U.S. jobs over the next four years, including electricians, machine technicians and indirect manufacturing positions.
Less than a week ago, the Wall Street Journal described Stargate as a "stalled" program. OpenAI CEO Sam Altman quickly refuted this statement on X, confirming that the total investment will "significantly exceed" the original promised $500 billion, and the number of GPUs is expected to exceed the 2 million mark this year.
Site selection for the additional 4.5 gigawatts of capacity remains under review, with Texas, Michigan, Wisconsin and Wyoming under consideration. Oracle has reportedly placed a $40 billion order for NVIDIA chips, underscoring the scale of its commitment. Electricity supply remains a key constraint, as 4.5 GW is equivalent to the average electricity consumption of 3.5 million households, necessitating simultaneous negotiations with utilities for on-site generation and dedicated transmission lines.
Abilene's H-shaped halls, designed to hold 50,000 GB200s each, are a live prototype for the campus and experience gained there, including liquid cooling loops and modular substations, will be replicated on future campuses.
